Orthogonal Projection from Point Cloud

Function Description

This step can convert a 3D point cloud into a 2D depth map through orthogonal projection, eliminating the “near-large-far-small” distortion caused by normal perspective projection, and ensuring the accuracy of geometric dimensions after projection.

Use Cases

This step is commonly used in measurement scenarios. By generating distortion-free depth maps and corresponding masks, it provides reliable input for high-precision measurement and 2D positioning.

Inputs

• Point Clouds with Normals: Original point cloud.

• Reference Poses: Reference pose that defines the projection coordinate system.

Outputs

• Mask: Binary mask image generated by projection.

• Image Center: Pixel coordinates aligned with the origin of the reference pose in the depth map.

• Resolution: Actual projection resolution used (unit: px/m).

• Depth: Image where pixel values represent the Z coordinate of points in the reference coordinate system.

Parameters

• Resolution Set Auto

  • Checked (Recommended): Automatically calculates the optimal resolution based on the point cloud range.

  • Unchecked: Need to manually specify Resolution.

• Resolution

  This parameter represents the number of pixels per unit length (1m) in the 2D image formed by orthogonal projection. Unit: px/m.

  Higher values mean higher precision but increased computation. It is recommended to adjust according to actual needs:

  • Adjust to an appropriate resolution to minimize holes in the depth map.

  • If time requirements are high, this parameter can be appropriately lowered.

• Mask Border Width

  The width of the border added around the projected mask image. Unit: pixels (px).

  This parameter can expand the range of ROI selection on the image and improve the robustness of subsequent processing. Recommended setting is 100px.